
Certifying neural networks for autonomous aircraft
Zurich-based Daedalean has written a 136 stage report with the European Union Aviation Safety Agency on assurance for the emergent properties of neural networks for autonomous aircraft.
“A lot of frameworks are written for traditional avionics. What I want from my operating system is that it doesn’t sit between me and the bare metal,” said Luuk Van Dijk, CEO of Swiss startup Daedalean.
The company is working with Honeywell on a joint testing and technological partnership for autonomous aircraft for takeoff, landing and GPS-independent navigation and collision avoidance in electric vertical takeoff and landing vehicles (eVTOL).
The report is at www.easa.europa.eu/document-library/general-publications/concepts-design-assurance-neural-networks-codann
Related articles
- SECURE-BY-DESIGN MULTI-CORE BOARD FOR INTEGRATED MODULAR AVIONICS
- SECURE KERNEL HYPERVISOR MOVES INTO ARM
- NEURAL COMPUTING ON A STICK: DEEP LEARNING HARDWARE BECOMES AFFORDABLE
